adapters. Also available for the IBM ServeRAID adapters are a range of battery backed cache options to provide even greater stability to your customer’s data.
5.2.5.6 RAID Summary
RAID should always be considered when availability and file size of business critical data is an issue. Individual requirements will dictate the final RAID solution for the customer.
In all cases there are advantages and disadvantages of each RAID solution in practical terms. These should be considered when selecting your hard disk configuration for the IBM Netfinity and J.D. Edwards’ OneWorld implementation.
Refer to Table 2 for an overview of advantages and disadvantages of RAID solutions. You should review 5.2.6, “Fault Tolerance” for further details regarding overall fault tolerance of the system.
Table 2. IBM Netfinity Hardware Considerations
Considerations | Advantages | Disadvantages |
|
|
|
RAID 0 | Very high read and write | Failure must be dealt with |
| performance | immediately with loss of access to |
|
| users |
|
| Data integrity - low |
|
|
|
RAID 1 | Data integrity - high | Degrades performance |
|
| (two writes for each data write) |
|
| Expensive, size of single disk is |
|
| limitation |
|
|
|
Disk Duplexing | Data integrity - very high | Single disk is size limitation |
| Performance is greater than RAID 1 | Expensive - 2x disk capacity for data |
| Higher fault tolerance than RAID 1 |
|
| Second disk controller |
|
|
|
|
Enhanced RAID 1 | Data integrity - very high | Performance impact |
| High fault tolerance | Performance - parity/checksum |
| No reboot on single disk failure | overhead |
|
|
|
RAID 5 | Data integrity - very high | Performance impact |
| High fault tolerance | Performance - parity/checksum |
| No reboot on single disk failure | overhead |
|
|
|
5.2.6 Fault Tolerance
As discussed in 5.2.5, “RAID Support” on page 39 there are limits to availability, and fault tolerance capabilities that various RAIDs can offer within the system by themselves. This section will look at other ways of offering fault tolerant solutions to get closer to the ultimate goal of 100% availability.
If we look at the most likely failures, given the history of IT technology, then we can start to look at the gain on reliability and availability versus cost of implementing the relevant solution. The most likely failures are discussed here in some order of "likely to fail", as this order can vary from country to country dependent upon many local factors.
Sizing, Considerations, and Recommendations | 43 |